The finished agreement is four or five tidy pages. Anyone can produce four tidy pages — a free template will do it for nothing, and an AI chatbot will do it in eleven seconds with impressive confidence. So it’s fair to ask what the $29 is for. The honest answer: the pages are the cheap part. What you’re paying for is the checking — a few dozen specific legal and mathematical checks, each one there because skipping it is exactly how family loans fall apart in a Canadian courtroom two years later. This guide lists them, plainly, so you can judge for yourself. (New here? Start with what LendRight is.)
The three ways people get four tidy pages
The US template. Most “free loan agreement” results are American: they cite state usury caps that don’t exist here, miss the Criminal Code entirely, say “State of ____”, and know nothing about a two-year limitation clock or an age of majority that’s 19 in six provinces and two territories. We’ve written before about what free templates miss; this guide is the deeper cut.
The AI chatbot. The text reads beautifully. But there is no canonical document — regenerate the conversation and the clauses drift; nobody can later prove which version was “the” agreement. The model happily writes “2% per month” (a phrasing the federal Interest Act punishes by capping recovery at 5% a year), invents notarization requirements Canada doesn’t have, assumes everyone is 18, and produces repayment tables whose arithmetic doesn’t survive a calculator. It also can’t tell you when the honest answer is this one needs a lawyer.
The fill-a-form site. Better than either — but the generic ones merge your names into fixed text. They don’t compute anything, don’t check anything against your dates or your provinces, and their clauses and their payment table were written by different people in different years.
What follows is the checklist LendRight actually runs. Each item names what goes wrong without it.
Check 1 — Was the money already sent?
Roughly half of family “loan agreements” are written after the money moved: the e-transfer went in March, the paperwork happens in July. Templates and chatbots paper this with a simple acknowledgement — “the Lender advanced $X and the Borrower agrees to repay.” Here’s the problem the common law has had for nearly two centuries: a promise given only in exchange for something that already happened can fail for lack of consideration. The borrower’s shiny new promises may be legally supported by nothing at all, leaving the lender to fall back on whatever fuzzy verbal arrangement existed before — minus every term in the document.
LendRight detects a funding date in the past and switches clause 2 into a different legal shape: the lender expressly agrees, in exchange for the borrower’s promises, to accept repayment on the schedule and not demand the money back while it’s being honoured — real, fresh consideration (the law calls it forbearance) — plus an express statement that the parties intend the agreement to be legally binding, which knocks out the “it was just family” presumption. And there’s a quiet bonus most people never learn until it’s too late: in most provinces you have roughly two years to sue once repayment fails, and a signed written acknowledgement of the debt restarts that clock (Ontario’s Limitations Act, 2002, s. 13 is the classic example). Papering an old loan properly doesn’t just document it — it can revive an enforcement window that was silently closing. A template doesn’t know the money already moved. A chatbot doesn’t ask.
Check 2 — The 35% line, and the leap-year trap inside it
Since January 1, 2025, charging an annual percentage rate above 35% is a criminal rate under s. 347 of the Criminal Code. Every serious tool blocks 35%. Here’s what almost nobody checks: interest in a proper agreement accrues daily, on the actual days elapsed, over a 365-day year — so across a leap year, a rate becomes rate × 366⁄365. A loan written at 34.99% quietly charges an effective 35.09% across 2028. That’s the kind of margin that only matters in the exact fight where everything matters.
So LendRight does two things templates do neither of. The builder blocks anything at 34.5% or above, keeping a deliberate buffer under the line even in leap years. And the agreement itself carries a safeguard clause — standard in professionally drafted Canadian credit documents, almost never in DIY ones — saying that if any amount would ever exceed the s. 347 maximum, it’s automatically reduced to the lawful ceiling and any excess goes against the principal. Courts have shown they’ll trim an offending rate rather than torch a loan when the document invites them to. Meanwhile the same screen catches the Interest Act s. 4 trap: any rate must be stated per annum, because “2% a month” without the yearly figure caps your recovery at 5% a year — a century-old rule AI reinvents its way into constantly. (Choosing a number? The interest guide and the calculator are for that.)
Check 3 — The words must match the math
Open a random template and read closely: the interest clause says “no compounding,” and three lines later interest accrues on any “overdue amount” — which includes overdue interest. That’s compounding. The contradiction sits there until a defaulting borrower’s lawyer reads both sentences aloud to a judge. AI is worse, because its payment table and its clauses are generated separately and agree only by luck: ask it for a schedule starting January 31 and watch the “February” payment land on March 3, because naive date math rolls the 31st forward.
In a LendRight agreement, one engine writes both the clauses and Schedule A. Interest accrues daily on the outstanding principal only — the clause says so, and the table does so. Payments apply to interest first, then principal. A January 31 anchor pays February 28 (or 29), and the clause discloses exactly that. The final row closes at exactly $0.00, to the cent, because the arithmetic is integer-cents all the way down. The words and the numbers cannot disagree, because they have the same author.
Check 4 — One document both people can prove
Here’s a failure mode nobody thinks about until it’s fatal: which document did you both agree to? A Word file gets edited after one person “agreed” by text message. An AI chat has no original at all. Even honest people end up with two slightly different PDFs and no way to say whose is real.
A LendRight agreement is sealed before anyone signs: the document date is frozen, the exact bytes are locked with a SHA-256 fingerprint, both signatures attach to those bytes, and the Certificate of Completion records the fingerprint, the timestamps, and each signer. Anyone, forever, can verify a copy against the fingerprint. There is exactly one agreement, and both people can prove which one it is.
Check 5 — Kindness isn’t a contract amendment
The single most common real-world defence in family lending has nothing to do with the original terms: “You took late payments for a year — you changed the deal,” or “Mum said at Christmas I could skip the winter.” Almost no template defends against it. LendRight’s agreement carries the two clauses that do: amendments count only in writing signed by both people, and accepting a late or partial payment — or being kind on one occasion — waives nothing and doesn’t rewrite the schedule. The lender gets to be generous without their generosity becoming a defence.
Check 6 — Drafted for humans, not corporations
Template default clauses are copy-pasted from commercial lending: “the Borrower’s reorganization,” “its bankruptcy” — corporate language for a loan between two people. Real family loans meet real human events, and the agreement should name them. LendRight’s does: a consumer proposal under the Bankruptcy and Insolvency Act — the single most common insolvency event for an individual Canadian, and the subject of its own guide here — is expressly a default event, so the entire balance crystallizes and the lender can file a proof of claim for the full amount, not just missed installments. The borrower’s death makes the debt a claim against the estate; the lender’s death (statistically likely somewhere in a ten-year parent-to-child loan) keeps everything payable to the lender’s estate on the same terms; incapacity hands the wheel to an attorney under a power of attorney without voiding anything. Templates go silent exactly where families need words.
Check 7 — Province-aware people checks
The age of majority is 18 in Ontario and Alberta and 19 in BC, Nova Scotia, New Brunswick, Newfoundland and Labrador, and all three territories — and it’s each signer’s own province that matters. LendRight checks each person’s date of birth against their own province’s age and writes clause 7 accordingly; a template assumes 18 and a chatbot doesn’t ask. Quebec is declined honestly (its civil law needs its own agreement, not a translated common-law one), and borrowers who live outside Canada are declined too — a document whose notice, capacity, and enforcement story assumes a Canadian borrower shouldn’t pretend otherwise. The notices clause points at email addresses that are actually printed in the document, deemed-receipt timing is spelled out, every payment carries a payment reference that turns the borrower’s own bank statements into a ledger for this exact loan, and a payment counts when it’s received — the detail that makes the grace-period clock unambiguous.
Check 8 — What we refuse to let you break
Some of the checking is refusal, and it’s worth paying for. Type “secured against the car” or “Dad is a guarantor” into the extras box and LendRight warns you on the spot: typed words create no enforceable security and bind no guarantor who hasn’t signed — that’s genuinely lawyer territory, and pretending otherwise would be selling you a false sense of cover. There are no hidden fees for the same reason: the agreement states that the APR is the rate, full stop, which is also what keeps the s. 347 arithmetic clean. And the document is deliberately a loan agreement rather than a promissory note, because classic negotiable instruments are carved out of some provincial e-signature statutes — the note-vs-agreement guide has the details.
The receipts
None of this is vibes. Every change to the agreement’s wording runs against hundreds of automated checks — including fully rendered sample agreements at different rates, payment frequencies, provinces, and funding dates — that verify the clauses, the schedule, the numbering, and the totals still agree, to the cent. The agreement also includes an independent-legal-advice acknowledgement, because the honest position is that a $29 document is the right tool for straightforward family loans and the wrong tool for secured, corporate, or estate-planning arrangements — here’s the line, plainly drawn.
| What gets checked | Free / US template | AI chatbot | LendRight |
|---|---|---|---|
| Money already advanced (consideration + limitation restart) | No | No | Yes — clause 2 switches shape |
| 35% criminal line incl. leap-year effect | Rarely (often US usury) | No — will draft 45% | Yes — 34.5% ceiling + safeguard clause |
| Rate stated per annum (Interest Act s. 4) | Sometimes | Often violated | Enforced |
| Clauses generated from the same engine as the schedule | No | No | Yes |
| Month-end dates (Jan 31 → Feb 28) | N/A — no schedule | Frequently wrong | Engine-clamped, clause-disclosed |
| One provable signed copy (fingerprint) | No | No original exists | SHA-256 sealed + verifiable |
| Late-payment tolerance ≠ waiver | Rare | Rare | Built in |
| Consumer proposal / lender death / incapacity | Corporate boilerplate | Silent | Named, in plain language |
| Per-person age of majority by province | No | No | Checked at entry |
| Tells you when you need a lawyer | No | No | Yes — and blocks the risky bits |
Frequently asked questions
Is an AI-written loan agreement enforceable in Canada?
It can be — a contract doesn’t care who typed it. The practical problems are everything around the text: no canonical signed copy anyone can prove, arithmetic that doesn’t match the clauses, US assumptions, rates phrased in ways the Interest Act punishes, and no check on ages, provinces, or the funding story. Enforceability fights are won on those details, not on fluent prose.
Why does LendRight cap interest at 34.5% when the legal line is 35%?
Because the agreement charges interest daily on actual days over a 365-day year, a nominal rate becomes rate × 366⁄365 across a leap year — 34.99% would drift to about 35.09%, over the criminal line. Capping at 34.5% keeps every year, including 2028, safely under it, and the agreement carries a safeguard clause on top.
What if I already sent the money months ago?
That’s one of the most important checks. LendRight rewrites the funding clause so the paperwork is supported by fresh consideration (the lender’s binding commitment to the schedule) and expressly intended to bind — and a signed acknowledgement of the debt can restart the limitation clock in most provinces. Don’t paper an old loan with a bare “we confirm this happened” template.
What does the $29 actually pay for?
The checking in this guide, the sealed and fingerprint-verifiable signing record, and the honesty — including being told when your situation needs a lawyer instead. Drafting is free; you only pay to finalize, and the person creating the agreement chooses who pays.
Related guides
How to lend money to family legally in Canada · Charging interest on a family loan · Real court cases on family loans · What is LendRight? · Do I need a lawyer?
Who runs this: LendRight is a product of RULE8 Inc. · Last reviewed July 6, 2026 by the LendRight Editorial Team. Based on: Criminal Code s. 347 (35% APR criminal rate, in force January 1, 2025); the federal Interest Act s. 4 (annual-rate disclosure); provincial limitation statutes (e.g. Ontario’s Limitations Act, 2002, s. 13 on acknowledgments); the Bankruptcy and Insolvency Act (consumer proposals; claims against estates); Pecore v. Pecore, 2007 SCC 17; and provincial electronic-commerce legislation. What this is (and isn’t): general information from a self-help drafting tool. LendRight is not a law firm; nothing here is legal advice and no lawyer-client relationship arises. Quebec’s civil-law regime differs and isn’t yet supported. Enforceability always depends on the facts, the parties, and the court.